Why are there two CN models, and what's the difference between them? (besides size)
I'm not sure why there are two models an which one to use.
Between v1 and v2, v2 is the better version, there is no reason to use v1 anymore.
If you mean between the .ckpt and the .safetensors, you should use the .safetensors by default (not just for this model, but in general, as .ckpt can contain malware), but from a function standpoint both are equivalent.
In the /v2 folder, there are 2 .safetensors:
control_v1p_sd15_qrcode_monster_v2.safetensors(and the.yaml) is compatible with AUTOMATIC1111's UIdiffusion_pytorch_model.safetensorsis compatible with thediffuserspython library, and thus also ComfyUI, so if that's what you're using, take this one
Thanks a lot, so /v2 .safetensors for A1111 is my go-to. Appreciate it!
Do we put the .yaml file in the same folder with the model - where all models for CN go?
Yes, you just put the .yaml along with the model.
I never quite know whether the "correct" place for CN models is in models\ControlNet, or extensions\sd-webui-controlnet\models.... ? (It doesn't seem to matter, but if one is the "old" way of doing it and is likely to be deprecated in the future, it would be useful to know!)
I don't have any more insight about this than you do, I always put it under extensions\sd-webui-controlnet\models and it works. I also haven't used sd-webui for a few months, so I'm not aware of any potential deprecation.